The Section has given special emphasis to the description and understanding of the neurobiology of the longitudinal course of affective disorders in light of the chronicity of the illness and its overwhelming proclivity for recurrence. The tendency for the frequency of cycling to accelerate, and the precipitation of episodes to become less dependent on psychosocial stresses over time, speaks to a potential sensitization process of prior episodes. This concept, pioneered in our Branch, has now been validated by the Denmark Patient Registry in 23,000 patients, demonstrating that the latency and incidence of recurrence of depressive episodes is directly proportional to the number of prior hospitalizations for depression in both unipolar and bipolar illness. New evidence of the pathological significance of episodes themselves in patients with affective disorder is derived from our observations that patients with a greater number of prior episodes of affective illness have increased dysfunction on a variety of neuropsychological tests. Our treatment-refractory affectively ill patients also have dysfunction in the recognition of facial emotional expression, as well as in navigation in geographic space, both of which are associated with neurophysiological abnormalities on positron emission tomography (PET) scans. Pre-clinical models for understanding molecular mechanisms involved in increased behavioral responsivity to the same stimulus over time have led to the postulate of the impact of stresses and episodes themselves on gene expression. This theoretical framework suggests that the cyclic presence or absence of affective dysfunction could be related to the relative ratio of pathological versus adaptive changes in gene expression. This model provides new targets for clinical study and therapeutics, not only in attempting to inhibit pathological changes, but also enhance endogenous adaptive mechanisms, such as the putative increases in TRH in depressive episodes. The positive antidepressant effects to intrathecal and parenteral TRH administration in depression provide preliminary confirmation of the hypothesis that the increases in TRH in depression could be a compensatory adaptation. In addition to the findings that some neuropeptides, such as somatostatin, are significantly low in the CSF of depressed patients in a state-dependent fashion, we have now obtained additional evidence of neuropeptide dysregulation in which significant peptide interrelationships that are normally observed in healthy control subjects are absent in our patient population, and vice-versa. In addition, we have continued to uncover heterogeneity of regional cerebral dysfunction in subgroups of affectively ill patients assessed with PET. Unipolar depressed patients show the classical picture of hypofrontality (correlated with severity on Hamilton depression ratings) compared with large age- and gender-matched groups of normal volunteers. Bipolar I patients tend to show the opposite pattern, with relative hypermetabolism, particularly in mediotemporal lobe structures. Bipolar II patients have greater variability and appear to represent a mixture of the hypo- and hypermetabolic subtypes, yielding generally nonsignificant differences from normal volunteer controls. The local anesthetic procaine has been found by PET to be a limbic-selective probe, and affectively ill patients are markedly hypoperfused in response to procaine compared with normal volunteers, suggesting substantial pathology in this limbic axis in depressed patients, as previously postulated. Preliminary data in primates implicate muscarinic m-2 receptors in procaine actions. The evidence of limbic system dysfunction in our patients, based on medication-free PET assessments in affectively ill patients at baseline, as well as PET studies in response to psychological probes (induction of happy, sad, angry, and anxious affects) and pharmacological probes (procaine), has led us to explore preclinical mechanisms of limbic dysfunction in vivo in studies of amygdala kindling and quenching, in collaboration with Susan Weiss, as well as in vitro in the amygdala slice preparation, in collaboration with He Li and Michael Rogawski's laboratory. These data have helped uncover the novel and frequency-dependent mechanisms for long-term changes in neuronal excitability that are of importance in their own right, but also helpful in generating a theoretical framework for differential frequencies of stimulation of affectively ill patients with repeated transcranial magnetic stimulation (rTMS) of the brain.